They probably want to hurry before it’s too late

Published May 27, 2020

German researchers say they’ve found a way to teach artificial intelligence right from wrong. Essentially, they “feed” it texts, and the machine learns to associate certain concepts as “good” or “evil.”

This allows the system to understand contextual information by analyzing entire sentences rather than specific words. As a result, the AI could work out that it was objectionable to kill living beings, but fine to just kill time.

Don’t be evil

The trick, of course, is to use the right texts. (Amazon learned this the hard way when its AI was downgrading graduates of all-women colleges.) The wrong ones “can lead to some extremely dubious ethical views, such as ranking eating animal products a more negative score than killing people.”

Tags: , , ,